An author has six or seven ideas about what might cause a financial accounting phenomena to be observed in markets.

All the six or seven ideas can be argued to be captured by one proxy variable, ABC.

The author proceeds to show that ABC is related to some capital markets construct of interest.

The author goes on the road with the goal of determining which of the six or seven ideas should be focused on as the “story” that goes with the empirical finding of a relationship.

WAIT!!!!  Is that not a form of data mining????

Unfortunately it appears today this is what passes for “successful” research in some areas of accounting. But you cannot tell this is what happened from the final journal article – indeed from reading the article one would think the paper based on this process was a fine case of positivist research at its best. Hypothesis development is presented as being done up front and proxies being picked based on that development.

I will explore what this means for us over the coming weeks as I think about this phenomena more.